Context Navigation

Changes between Initial Version and Version 1 of Ticket #1135, comment 4

-                      initial
+                      v1
 * the "backlog-factor" measures how many pixels and packets should have been ACKed by the client but haven't - this should almost never happen and we have code that takes immediate action for this particular problem, the "backlog-factor" just exposes it in the more usual place (I have edited comment:1 to fix the documentation of the data format)
 * the "compression-ratio" uses the format {{{AVERAGE, RECENT, DELTA}}} - the delta value tells you in which direction we adjust the quality (ie: -10 means we drop the quality by 10 points)
+* the "compression-ratio" uses the format {{{AVERAGE, RECENT, DELTA}}} - the delta value tells you in which direction we adjust the quality (ie: -10 means we drop the quality by 10 points). If things are better recently then we raise the quality, but if they're worse then we lower it.
 Based on the data above, and some quick tests, I have tweaked the heuristics a bit:
 * r13589 changes in compression ratio should not sway the quality so much
+* r13589 changes in compression ratio should not sway the quality quite so much
 * r13590 allow the quality to go higher for "reasonable" values of the batch delay